Skip to content
How to ThinkIn the Age of AI
Rulev2RL-873

Limit checklists to 5-9 items targeting frequent +…

Limit output checklists to five-to-nine items targeting errors that are both frequent and consequential, ordering from most catastrophic to least to enable partial execution under time pressure.

Why This Is a Rule

Atul Gawande's The Checklist Manifesto demonstrated that checklists dramatically reduce error rates — but only when they're short enough to actually use. A 20-item checklist takes so long to complete that people skip it under time pressure, producing zero error reduction when errors are most likely (during rushes and fatigue). A 5-9 item checklist takes 1-3 minutes and survives time pressure because the effort is small enough to always execute.

The 5-9 range maps to working memory capacity: you can hold 5-9 items in mind as a coherent set without losing track of where you are. Below 5, the checklist may miss important errors. Above 9, the checklist itself becomes cognitively demanding, and compliance drops because checking feels like work rather than a quick verification.

The frequent AND consequential filter ensures every item earns its place. An error that's frequent but inconsequential (typo in an internal chat) doesn't need a checklist item. An error that's consequential but rare (sent to wrong CEO — once in your career) doesn't justify a permanent spot. Only errors that are both common enough to occur regularly and serious enough to matter when they do deserve checklist real estate.

The catastrophic-first ordering enables graceful degradation: when you're truly rushed and can only check 3 of 7 items, the first 3 catch the most catastrophic errors. The checklist is front-loaded so that partial execution catches the worst failures.

When This Fires

Common Failure Mode

The comprehensive checklist: including every possible error, producing a 25-item list that takes 15 minutes to complete. Compliance drops to 20% because the list is too long for routine use. A 7-item list with 95% compliance catches more errors than a 25-item list with 20% compliance.

The Protocol

(1) List all errors that have occurred in this output type over the past 6 months (Build checklists from documented error history…). (2) Score each error on two dimensions: Frequency (how often does this occur?) and Consequence (how bad is it when it does?). (3) Select the 5-9 errors that score high on both dimensions. These are your checklist items. (4) Order from most catastrophic to least: the first item should catch the error with the worst consequences. (5) Test the checklist: can you complete it in under 3 minutes? If not, reduce items until you can. Speed of completion determines compliance, and compliance determines error reduction.